Container Resource Management via Usage Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing resource management methods fail to effectively determine container allocation in nodes based on varying resource usage patterns, leading to potential resource surplus or underutilization, as they do not account for peak times and usage variations among containers.
Innovation Solution
A resource management device that acquires monitoring data, predicts future resource usage patterns, and dynamically adjusts container allocation by deleting containers when predicted total usage exceeds available resources, ensuring resource utilization remains within node limits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If containers are determined on the basis of respective peak values of resource usage, then resource allocation is simplified, but a large portion of node resources remains unused
Solution Approach 1:
The patent performs preliminary actions by acquiring historical resource usage data and predicting future usage patterns before making container allocation decisions. The prediction unit forecasts resource usage for each container at future time points, and the total resource usage is calculated in advance to determine which containers can be mounted, preventing both resource surplus and overload situations before they occur.
2Adaptability or versatility
If containers are determined on the basis of resource usage at a freely-selected time, then resource allocation is flexible, but total resource usage may exceed node capacity in the future
Solution Approach 1:
The patent implements feedback mechanisms by continuously monitoring actual resource usage against predicted usage patterns. The system acquires historical usage data, predicts future usage, makes allocation decisions, and can re-evaluate by acquiring new monitoring data to update predictions, ensuring that resource capacity constraints are maintained while allowing flexible container management.
3Device complexity
If traditional resource management methods are used without considering variation states, then management complexity is reduced, but resource surplus or overload situations occur
Solution Approach 1:
The patent applies parameter changes by transitioning from static resource allocation based on peak values or arbitrary time points to dynamic allocation based on predicted resource usage parameters. The system calculates total resource usage by summing predicted usage across all containers at each future time point, enabling accurate resource utilization management that adapts to varying container resource patterns.
Data Source
AI summary
A resource management device includes a data acquiring unit to acquire monitoring data indicating a temporal change in a use amount of a resource used in the past by each of containers included in a node among resources of the node, a prediction unit to predict a temporal change in a use amount of the resource to be used in the future by the container from each of pieces of the monitoring data, and calculate a temporal change in a total use amount of the resource to be used in the future by the containers from the prediction result, and a container deleting unit to delete one of the containers when a timing at which the total use amount of the resource becomes larger than the total of the resources of the node is included in the temporal change in the total use amount of the resource.